Automated outbreak detection systems in the EU: Requirements and challenges for its implementation, 2023/2024

This paper outlines the current status, prerequisites, and implementation challenges of automated outbreak detection systems across the EU and UK, revealing that while demand and data availability are high, adoption is hindered by funding and IT constraints, suggesting that EU-funded collaborative capacity building could enable the development of a shared, sustainable solution.

The Gist

Imagine the European Union as a massive, bustling city with 27 different neighborhoods (countries), each watching over its own health. The goal is to spot a "fire" (an infectious disease outbreak) the moment a single spark appears, so firefighters can rush in before the whole building burns down.

This paper is like a city-wide inspection report asking: "Who has the best fire alarms? Who is still using a smoke detector from 1990? And why aren't more people installing new ones?"

Here is the breakdown of the report in simple terms:

1. The Big Idea: Automatic Fire Alarms

In the past, health officials had to manually check the daily newspaper of disease cases, looking for weird spikes. It was like a security guard watching 100 CCTV cameras with their eyes, hoping to spot a thief.

Automated Outbreak Detection Systems (AODS) are like smart, AI-powered fire alarms. They constantly watch the data, compare it to what usually happens (the "baseline"), and instantly scream, "Hey! Something is wrong here!" if the numbers jump too high. This gives health officials a head start to stop diseases before they spread.

2. The Current Situation: A Mixed Bag

The researchers asked 21 countries in the EU and the UK, "Do you have these smart alarms?"

• The Good News: Seven countries (like Germany, the UK, and the Netherlands) already have working systems. They are the "smart cities" of the group.
• The Bad News: Most other countries are still doing it the old-fashioned way or don't have the tools at all.
• The Trend: Everyone wants the smart alarms. In fact, 10 countries are currently testing a new, shared alarm system developed by a joint EU project called "United4Surveillance."

3. The Problem: Why Isn't Everyone Using Them?

If the alarms are so great, why isn't everyone buying one? The paper found three main reasons, which we can think of as The Three Leaks in the Bucket:

• Leak #1: No Money (The Empty Wallet): This is the biggest problem. Setting up these systems costs cash. Countries say, "We have the data, but we don't have the budget to hire the programmers or buy the servers." It's like having a Ferrari but no gas money.
• Leak #2: No Mechanics (The Missing Skills): Even if you have the money, you need people who know how to build and fix the alarm. Many countries lack experts in data science and IT. They have the raw ingredients (data) but no chef to cook the meal.
• Leak #3: The Data is Messy (The Broken Pipes): Sometimes the data coming in is late, incomplete, or in a format the alarm can't read. It's like trying to run a high-tech alarm on a house with rusty, leaking pipes.

4. The Solution: One Tool to Rule Them All?

Here is the most exciting part of the paper. The researchers found that even though every country is different, they all want the exact same thing.

• They all want to see the data on a map.
• They all want to filter by age or region.
• They all want a simple "red flag" when something is wrong.

Because the needs are so similar, the authors suggest we don't need to build 27 different alarms. Instead, we can build one super-flexible, open-source "Swiss Army Knife" tool that every country can use. It's like giving every neighborhood the same high-quality fire truck, just with a different paint job for their local flag.

5. The Future: Teamwork Makes the Dream Work

The paper concludes that while money and tech skills are holding people back, the EU is stepping in to help. Through projects like "United4Surveillance," they are:

• Building that shared "Swiss Army Knife" tool.
• Training people on how to use it.
• Helping countries connect their data pipes so the alarm works smoothly.

In a nutshell: The technology to spot disease outbreaks instantly exists and works well. The only thing stopping us from saving lives faster is a lack of funding and technical know-how. If the EU countries pool their resources and use a shared tool, they can turn the whole continent into a city with the best fire alarm system in the world.

Technical Summary

1. Problem Statement

Timely detection of infectious disease outbreaks is critical for public health surveillance, yet information regarding the implementation of Automated Outbreak Detection Systems (AODS) across the European Union (EU) and the United Kingdom (UK) remains limited and fragmented. While statistical methods for detecting anomalies (e.g., Farrington, Moving Epidemic Method) are well-established, there is a lack of updated data on:

• The current adoption rate of AODS among National Public Health Institutes (NPHI).
• The specific prerequisites (data quality, IT infrastructure) required for implementation.
• The primary barriers (funding, technical capacity) preventing wider adoption.
• The convergence of user needs across different national contexts, which could inform the development of shared, scalable tools.

2. Methodology

The study employed a mixed-methods approach as part of the EU4Health Joint Action UNITED4Surveillance (U4S) to assess the landscape of AODS in 26 countries (25 EU Member States + UK).

• Data Collection:
  • Survey: An online questionnaire distributed to 25 countries in April 2023. It contained 33 questions (22 mandatory) covering surveillance system characteristics (data sources, aggregation, frequency) and AODS capacities (methods, resources, outputs). 21 countries (84% response rate) completed the survey.
  • Workshop: An in-person workshop in May 2023 with 30 experts from 16 countries to validate survey data, discuss use cases, and define tool requirements.
  • Case Studies & Meetings: In-depth analysis of the UK and France (based on literature and prior knowledge) and one-on-one meetings with 10 pilot countries in early 2024 to gather specific functional requirements.
• Data Analysis: Descriptive analysis of survey data using R (v4.4.0). Results were validated by participants during the workshop and subsequent webinars.
• Scope: The analysis focused on three objectives: assessing demand, examining data prerequisites, and identifying implementation challenges.

3. Key Contributions

• Updated Landscape Assessment: Provides the first comprehensive overview of AODS status in the EU/UK since 2010, identifying that only 7 countries currently have operational AODS, while 10 are piloting new tools and 4 are planning implementation.
• Requirement Harmonization: Demonstrates that despite diverse national surveillance structures and statistical methods, the functional requirements for AODS (e.g., flexible outputs, stratification options, user-friendly interfaces) are highly convergent.
• Gap Analysis: Systematically identifies the disconnect between the availability of high-quality surveillance data and the lack of technical/financial resources to automate analysis.
• Tool Development Roadmap: Defines specific technical specifications for a shared, open-source AODS tool (R Shiny app) capable of serving multiple jurisdictions.

4. Key Results

A. Current Implementation Status

• Active Systems: Seven countries (Austria, Denmark, Finland, Germany, Netherlands, France, UK) have established AODS.
• Methodological Diversity:
  • Statistical Methods: Most systems use FarringtonFlexible (Austria, Denmark, Germany, UK Lab), Moving Epidemic Method (MEM) (Finland), Hidden Markov Models (Netherlands, France), or RAMMIE (UK Syndromic).
  • Data Sources: Ranges from specific pathogen lab data (e.g., Campylobacter in Austria) to comprehensive syndromic surveillance (e.g., ED visits, GP consultations in France and UK).
  • Automation Levels: Vary from fully automated weekly reports (Germany, Austria) to interactive dashboards requiring ad-hoc computation (Netherlands).
• Pilot Expansion: Interest is high; the number of countries piloting the U4S-developed tool grew from 7 to 10.

B. Prerequisites and Data Availability

• Data Quality: The majority of countries possess the necessary data infrastructure:
  • 20/21 report case-based data daily.
  • 21/21 have at least 3 years of historical data (8 have >20 years).
  • 21/21 receive stratification by age, region, pathogen, and sex.
  • Many have additional data on hospitalization, death, and vaccination status, enabling complex signal analysis.

C. Barriers to Implementation

Despite high demand and data availability, adoption is stifled by:

1. Resource Constraints (Primary Barrier): 15 countries cited a lack of sustainable funding and resources. Specific deficits include:
   • Research capacity and staff time (n=12).
   • Data science expertise (n=8).
   • IT support and maintenance (n=6).
2. Data Quality Issues: Reporting delays and limited data availability (n=9).
3. Technical/Legal Hurdles: System incompatibilities during upgrades (e.g., Sweden's system discontinuation) and legal mandates (n=5).

D. Identified Use Cases and Requirements

• Priority Pathogens: Gastroenteritis (food/waterborne) and respiratory diseases are the primary use cases.
• Functional Needs: Users require flexible output formats, multi-language support, temporal aggregation (daily/weekly/monthly), and the ability to filter by demographics and known outbreaks.
• User Profile: Systems must accommodate users with varying levels of IT expertise, necessitating intuitive, "user-friendly" interfaces.

5. Significance and Future Perspectives

• Standardization Potential: The study concludes that a single, flexible, open-source tool (such as the R Shiny app developed in the U4S project) can serve multiple countries. The convergence of outputs (signals, maps, reports) across different statistical methods suggests that standardization is feasible without sacrificing national specificity.
• Capacity Building: EU-funded Joint Actions are critical for bridging the gap between data availability and technical implementation by fostering cross-country collaboration and providing sustainable tools.
• Future Challenges:
  • Signal Overload: Automated systems generate high volumes of signals, requiring systematic evaluation to manage verification workloads and avoid alert fatigue.
  • New Data Streams: Future AODS must integrate Whole Genome Sequencing (WGS) data, requiring new interoperability standards.
  • Sustainability: Long-term success depends on securing dedicated funding for software engineering and maintenance, moving beyond project-based funding.

In summary, the paper argues that while the technical and data foundations for automated outbreak detection exist across the EU, the bottleneck is human and financial capital. Addressing these resource gaps through collaborative, shared tools is the most viable path to enhancing early warning systems for infectious diseases in Europe.